[Catherine Lai]

Hi Alex,

As far as I can tell, you can still submit after the deadline as long as you have started before. But to be honest, when it comes to things on Learn it’s better to be safe than sorry! I would aim to submit before 12 noon, if possible.

cheers,
Catherine

[Catherine Lai]

Also, I’ve redacted the quiz questions posted above as this is a timed test. Please don’t post test questions on the forum!

[Catherine Lai]

Hi George (and everyone)

Sorry, about the confusion here. It’s unexpected behaviour from Learn. However, you can take that behaviour as a (unplanned) hint from Learn. Only selecting one option for those questions won’t hurt your grade!

cheers,
Catherine

[Catherine Lai]

Hi Xueyan,

You’re right, the lowest analysis frequency is DFT[1] (i.e., k=1), but the actual DFT outputs start at index k=0. We skipped over a bit in class, but essentially the DFT[0] represents the vertical “bias” of the waveform. This tells you whether the waveform in the input window is symmetric around 0 (in amplitude) or whether it’s shifted up (or down on average.

This bias terms doesn’t actually tell us what frequencies are present in the input, so we usually don’t think too much about it. But, you can see a difference in DFT[0] if you compare a sine wave that matches one of the analysis frequencies as input (which has amplitude ranging from -1 to +1) and an input window containing one impulse and the rest zeros (so has amplitude either 0 and +1). In this case, DFT[0] for the sine wave will be zero, while DFT[0] for the impulse will 1.

In terms of the mirroring, DFT[0] would mirror DFT[16] where the number of samples in the input window is N=16. But the DFT formulation doesn’t include this (k=0,…,N-1). But, like DFT[0], DFT[N] wouldn’t tell us anything about the frequency components in the input.

cheers,
Catherine

[Catherine Lai]

Hi Ha Anh,

This spectrogram was actually created by Rebekka, but from the general spectral characteristic, I’d say:
(1) is a fricative, (4) is a vowel-approximant-vowel sequence,
(10) is a vowel with glottalisation at the end (11) is probably an affricate (i.e. a plosive + fricative).

But I’ll have to check with Rebekka what they actually are!

We didn’t do much spectrogram reading in this course, so we’d only ask you to recognise broad categories of speech sounds (with more easily recognisable acoustic features!).

cheers,
Catherine

[Catherine Lai]

Hi Antonia,

The updated version should be ready for you next week, but note that we’ll officially start working on assignment one in the module 5 lab (so in week 6).

cheers,
Catherine

[Catherine Lai]

Hi Yujia,

The way we talked about it in class,
X(m) represents the magnitude spectrum of the input window x(n). Similarly, H(m) represents the magnitude spectrum of the filter h(k).

So if x(n) represents the impulse train, X(m) will represent the harmonics.
Since h(k) represents the filter, H(m) will give us the shape of the spectral envelope.

But it’s worth noting that the convolution theorem also applies more generally, x(n) and h(k) could represent other types of signals (rather than just source and filter). But we are focusing on source and filter in this class.

cheers,
Catherine

[Catherine Lai]

Hi Tom,

I can connect to that machine from my end. Can you check whether you have the university VPN on (if you are outside the campus eduroam network)?If this happens on other machines as well,it would be best to contact IS Helpline to see if there’s anything the sys admins can see on their end:

https://www.ishelpline.ed.ac.uk/forms/

For VNC Viewer, you shouldn’t need an account with RealVNC (just the EASE login and password to access the machine).

cheers,
Catherine

[Catherine Lai]

Hi Eilish,

I think you might be on the Informatics remote desktop (since the computer name on your terminal is “theorbo” (which is an informatics machine). You can try the command hostname in the terminal to check if it has an inf.ed.ac.uk name.

From what I remember, the Informatics servers do have Festival installed, but it’s an older version than the one we are using, and the voice data is not there.

The remote desktop we are using is a separate service run by PPLS (cross-school IT issues, sorry!) which you will need to connect to separately. To connect to the PPLS AT lab remote desktop, you’ll need to follow the instructions here: https://speech.zone/courses/speech-processing/module-0-getting-started/ (see Computing Requirements) and pick one of the machines from this list:

Module 0 – getting started

You’ll need to have the University VPN on for this to work (though the Informatics specific one might work as well).

cheers,
Catherine

[Catherine Lai]

I’ve checked that it should be readable to you if you can log into the machine. Can you tell me what the output is if you run the following command:

ls -lh /Volumes/Network/courses/sp/assignment1/

[Catherine Lai]

Hi Eilish,

Would you be able to put this in as help request on via IS Helpline?

https://www.ishelpline.ed.ac.uk/forms/

The lab system administrators should be able to help. I’ll also ask them, but they can give you more specific help for your setup.

cheers,
Catherine

[Catherine Lai]

Hi Nickon,

For the course intro and signals lectures, you can download pdfs of the google slides from the shared links (File > Download > PDF document)

You can find the links at the bottom of the “Start” tab of the modules or here:

Module 1

Module 3

Module 4

I’ll add the pdfs to the Module start pages, but if you want them right away you can get them directly via those links.

I don’t think Rebekka had specific slides for the module 2 lecture.

cheers,
Catherine

[Catherine Lai]

Hi Muminah, everyone

I’ve put solutions on Learn: practice question solutions pdf

On learn: Assessment > Practice Questions > Practice Question Solutions

cheers,
Catherine

[Catherine Lai]

Hi Paige,

You’re not missing anything! The thing to take away is that the difference between cosine and sine waves of the same frequency (and amplitude) can be characterised by a 90 degree (pi/2) phase shift.

The idea was to either try different phase shifts until you get something that looks like a cosine wave or to derive this based on the fact that one period is equivalent to 2pi (and you know cosine start at the positive peak of the wave so 1/4 into the sine wave cycle). But, yes, you might just know it as a mathematical identity!

cheers,
Catherine

[Catherine Lai]

Hi Christina,

It’s a complicated issue, but there’s some discussion of this in this public facebook group that you might find interesting:
https://www.facebook.com/groups/551898862199586/permalink/695289281193876

One of the issues is that contrasts that are important for many non-European languages are relegated to diacritics or general made to seem much less frequent than they actually are if we look at different languages and speakers. Basically it’s not a theory neutral representation of the range of speech sounds.

There’s a nice interface that links to papers that describe how to transcribe non-European languages here that you might find interesting (noted by Amalia Arvaniti in that thread).

Similarly, many speech technologies depend on phonetic transcription (e.g. pronunciation dictionaries), where the transcription system is tailored to the language or accent that you want to develop the technology for. So in practice the IPA doesn’t work that well as a sort of general purpose transcription system (though there are some other computational issues that historically came into play here!).

Rebekka will have more to say about this but it may also be something to discuss in office hours (as Rebekka probably won’t be in the lab again for another few weeks).

cheers,
Catherine

[Author]

Posts